On 6 Jun 2002 11:39:37 -0700, [EMAIL PROTECTED] (Wahab El-Naggar) wrote: > > Dear Colleagues > > Thank you for your response about my questions with, especial appreciation > to Rich Ulrich, Jay Warner, and Charming. > Here are some details about my experiment: > It is a biomechanics study. The subjects ran in two different situations, > bear foot and with shoes, and we measures 18 angles of different parts of > the body to examine the differences in these angles in the two situations. > The following are the first 18 measures, followed by the second 18 measures > and data sample of three subjects. > My question again is what is the most suitable analysis to see the > differences in the 18 measurements in the two situations. > ... and ... you are into this without hypotheses? without knowledge of what *ought* to be? An "informed" analysis has a huge head-start on one that starts from total ignorance.
If you are seriously interested in "differences," then the sensible thing would seem to be: - Subtract one from the other. Whatever is reliably different from 0 is different. If this is totally exploratory, I guess you look at each variable separately, and each pair. And you look to see where it makes sense to combine two measures -- I mean, ask the experimenters, and look into the physical quantities being measured. If angle C= (A)+ (B) , does (C) make sense by itself? Or, might you combine angles using lessons from geometry? If it is not entirely exploratory, then you have prior results to work from; the task is to see what new information might add to what you knew before. -- Rich Ulrich, [EMAIL PROTECTED] http://www.pitt.edu/~wpilib/index.html . . ================================================================= Instructions for joining and leaving this list, remarks about the problem of INAPPROPRIATE MESSAGES, and archives are available at: . http://jse.stat.ncsu.edu/ . =================================================================
